Epub: only attach images referenced in the text.

[librarian.git] / librarian / book2anything.py

diff --git a/librarian/book2anything.py b/librarian/book2anything.py
index 7ae6178..948d9fd 100755 (executable)
--- a/librarian/book2anything.py
+++ b/librarian/book2anything.py
@@ -4,13 +4,14 @@
 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #
 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #

-from collections import namedtuple
+from __future__ import print_function, unicode_literals
+

 import os.path
 import optparse
 import os.path
 import optparse

-
+import six

 from librarian import DirDocProvider, ParseError
 from librarian.parser import WLDocument
 from librarian import DirDocProvider, ParseError
 from librarian.parser import WLDocument

-from librarian.cover import DefaultEbookCover
+from librarian.cover import make_cover

 
 
 class Option(object):
 
 
 class Option(object):


@@ -34,16 +35,15 @@ class Book2Anything(object):
     
     Subclass it for any format you want to convert to.
     """
     
     Subclass it for any format you want to convert to.
     """

-    format_name = None # Set format name, like "PDF".
-    ext = None # Set file extension, like "pdf".
-    uses_cover = False # Can it add a cover?
-    cover_optional = True # Only relevant if uses_cover
-    uses_provider = False # Does it need a DocProvider?
-    transform = None # Transform method. Uses WLDocument.as_{ext} by default.
-    parser_options = [] # List of Option objects for additional parser args.
-    transform_options = [] # List of Option objects for additional transform args.
-    transform_flags = [] # List of Option objects for supported transform flags.
-
+    format_name = None  # Set format name, like "PDF".
+    ext = None  # Set file extension, like "pdf".
+    uses_cover = False  # Can it add a cover?
+    cover_optional = True  # Only relevant if uses_cover
+    uses_provider = False  # Does it need a DocProvider?
+    transform = None  # Transform method. Uses WLDocument.as_{ext} by default.
+    parser_options = []  # List of Option objects for additional parser args.
+    transform_options = []  # List of Option objects for additional transform args.
+    transform_flags = []  # List of Option objects for supported transform flags.

 
     @classmethod
     def run(cls):
 
     @classmethod
     def run(cls):


@@ -53,33 +53,21 @@ class Book2Anything(object):
 
         parser = optparse.OptionParser(usage=usage)
 
 
         parser = optparse.OptionParser(usage=usage)
 

-        parser.add_option('-v', '--verbose', 
-                action='store_true', dest='verbose', default=False,
-                help='print status messages to stdout')
-        parser.add_option('-t', '--html-toc', 
-                action='store_true', dest='html_toc', default=False,
-                help='with inline html toc [book2epub only]')
-        parser.add_option('-k', '--use-kindlegen', 
-                action='store_true', dest='use_kindlegen', default=False,
-                help='use kindlegen tool [book2mobi only]')
-        parser.add_option('-d', '--make-dir',
-                action='store_true', dest='make_dir', default=False,
-                help='create a directory for author and put the output file in it')
-        parser.add_option('-o', '--output-file',
-                dest='output_file', metavar='FILE',
-                help='specifies the output file')
-        parser.add_option('-O', '--output-dir',
-                dest='output_dir', metavar='DIR',
-                help='specifies the directory for output')
+        parser.add_option('-v', '--verbose', action='store_true', dest='verbose', default=False,
+                          help='print status messages to stdout')
+        parser.add_option('-d', '--make-dir', action='store_true', dest='make_dir', default=False,
+                          help='create a directory for author and put the output file in it')
+        parser.add_option('-o', '--output-file', dest='output_file', metavar='FILE',
+                          help='specifies the output file')
+        parser.add_option('-O', '--output-dir', dest='output_dir', metavar='DIR',
+                          help='specifies the directory for output')

         if cls.uses_cover:
             if cls.cover_optional:
         if cls.uses_cover:
             if cls.cover_optional:

-                parser.add_option('-c', '--with-cover', 
-                        action='store_true', dest='with_cover', default=False,
-                        help='create default cover')
-            parser.add_option('-C', '--image-cache',
-                    dest='image_cache', metavar='URL',
-                    help='prefix for image download cache' +
-                        (' (implies --with-cover)' if cls.cover_optional else ''))
+                parser.add_option('-c', '--with-cover', action='store_true', dest='with_cover', default=False,
+                                  help='create default cover')
+            parser.add_option('-C', '--image-cache', dest='image_cache', metavar='URL',
+                              help='prefix for image download cache' +
+                              (' (implies --with-cover)' if cls.cover_optional else ''))

         for option in cls.parser_options + cls.transform_options + cls.transform_flags:
             option.add(parser)
 
         for option in cls.parser_options + cls.transform_options + cls.transform_flags:
             option.add(parser)
 


@@ -87,7 +75,7 @@ class Book2Anything(object):
 
         if len(input_filenames) < 1:
             parser.print_help()
 
         if len(input_filenames) < 1:
             parser.print_help()

-            return(1)
+            return 1

 
         # Prepare additional args for parser.
         parser_args = {}
 
         # Prepare additional args for parser.
         parser_args = {}


@@ -98,31 +86,28 @@ class Book2Anything(object):
         for option in cls.transform_options:
             transform_args[option.name()] = option.value(options)
         # Add flags to transform_args, if any.
         for option in cls.transform_options:
             transform_args[option.name()] = option.value(options)
         # Add flags to transform_args, if any.

-        transform_flags = [flag.name() for flag in cls.transform_flags
-                    if flag.value(options)]
+        transform_flags = [flag.name() for flag in cls.transform_flags if flag.value(options)]

         if transform_flags:
             transform_args['flags'] = transform_flags
         if options.verbose:
         if transform_flags:
             transform_args['flags'] = transform_flags
         if options.verbose:

-            transform_args['verbose'] = True        
-        if options.html_toc and cls.ext == 'epub':
-            transform_args['html_toc'] = True
-        if options.use_kindlegen and cls.ext == 'mobi':
-            transform_args['use_kindlegen'] = True
+            transform_args['verbose'] = True

         # Add cover support, if any.
         if cls.uses_cover:
             if options.image_cache:
         # Add cover support, if any.
         if cls.uses_cover:
             if options.image_cache:

-                def cover_class(*args, **kwargs):
-                    return DefaultEbookCover(image_cache=options.image_cache, *args, **kwargs)
+                def cover_class(book_info, *args, **kwargs):
+                    return make_cover(book_info, image_cache=options.image_cache, *args, **kwargs)

                 transform_args['cover'] = cover_class
             elif not cls.cover_optional or options.with_cover:
                 transform_args['cover'] = cover_class
             elif not cls.cover_optional or options.with_cover:

-                transform_args['cover'] = DefaultEbookCover
-
+                transform_args['cover'] = make_cover

 
         # Do some real work
         try:
             for main_input in input_filenames:
                 if options.verbose:
 
         # Do some real work
         try:
             for main_input in input_filenames:
                 if options.verbose:

-                    print main_input
+                    print(main_input)
+
+            if isinstance(main_input, six.binary_type):
+                main_input = main_input.decode('utf-8')

 
             # Where to find input?
             if cls.uses_provider:
 
             # Where to find input?
             if cls.uses_provider:


@@ -135,7 +120,7 @@ class Book2Anything(object):
             if not (options.output_file or options.output_dir):
                 output_file = os.path.splitext(main_input)[0] + '.' + cls.ext
             else:
             if not (options.output_file or options.output_dir):
                 output_file = os.path.splitext(main_input)[0] + '.' + cls.ext
             else:

-                output_file = None
+                output_file = options.output_file

 
             # Do the transformation.
             doc = WLDocument.from_file(main_input, provider=provider, **parser_args)
 
             # Do the transformation.
             doc = WLDocument.from_file(main_input, provider=provider, **parser_args)


@@ -144,12 +129,11 @@ class Book2Anything(object):
                 transform = getattr(WLDocument, 'as_%s' % cls.ext)
             output = transform(doc, **transform_args)
 
                 transform = getattr(WLDocument, 'as_%s' % cls.ext)
             output = transform(doc, **transform_args)
 

-            doc.save_output_file(output,
-                output_file, options.output_dir, options.make_dir, cls.ext)
+            doc.save_output_file(output, output_file, options.output_dir, options.make_dir, cls.ext)

 
 

-        except ParseError, e:
-            print '%(file)s:%(name)s:%(message)s' % {
+        except ParseError as e:
+            print('%(file)s:%(name)s:%(message)s' % {

                 'file': main_input,
                 'name': e.__class__.__name__,
                 'message': e
                 'file': main_input,
                 'name': e.__class__.__name__,
                 'message': e

-            }
+            })